Creative system

The creative system is the thing the positioning says the operator built. These are its actual working rules, taken from how Eddie teaches it on recorded calls. verified as spoken; the system itself evolves

Branded Response Flywheel#

The company's creative process has five stages:

1. Deep AI and competitor research
2. Performance creative strategy
3. Iterative creative formula
4. Testing and intelligence extraction
5. Profit-driven scaling

Research supplies the raw concepts. Strategy decides which concepts deserve the first tests. The iterative formula expands them into hooks and formats. Testing records what each combination teaches. Profit-driven scaling directs money toward validated branches, and the resulting intelligence feeds the next round of research and strategy. verified by Eddie

The Creative Tree#

The Creative Tree is the flywheel's memory. Each concept is a trunk. Hooks and formats form its branches. Every combination is marked as validated, unsuccessful, or untested.

The tree becomes active during the Iterative Creative Formula: Concept → Hook → Format. Testing and Intelligence Extraction writes the result back to the relevant branch. Profit-Driven Scaling uses that map to decide where the next pound should go.

The tree solves a practical constraint: no brand can test every concept, hook, and format combination in parallel without unlimited budget. A visible history lets the team find patterns across concepts, bring a proven format into a promising untested branch, and avoid repeating work it has already done. verified by Eddie; tree metaphor recorded in 2026-03-31 team-desiree

The operating point is: "If your creative process has no memory, every month starts from zero."

The DNA hierarchy: idea, hook, format#

Every ad decomposes into three components, in order of leverage: the idea (which persona or problem the creative calls out; "the idea will have the biggest pull... if the idea is shit, the other two could maybe carry it a little bit, but it's just going to die"), the hook (the first three seconds), and the format (founder video, UGC, static, POV, news-style, animated static). Targeting happens in the creative, not in the audience settings: "instead of the audience targeting... we're actually approaching the creative with that in mind, trying to call out the audience we're trying to hit" (2026-07-22 peer-antonio).

Diversity is a strategy, not a preference#

Different formats land with different people, and Meta throttles sameness: "if all of our creatives look like that, Facebook will think they're all the same creative and it won't show them to more people" (2026-05-06 client-vibae-monthly). Problem-first ads convert best right now, followed by solution framing, with lifestyle and mood as the long tail (2026-07-01 client-vibae-monthly). A good ad clears three bars: elicits an emotional response, is instantly understood, creates urgency; anything that clears none is "a filler ad" (2026-07-31 team-creative-planning).

Measurement is built into the naming#

Every creative's name encodes its idea, hook, and format plus production metadata (length, captions, voice type, setting), so performance can be decomposed later: "give me everything with this hook, give me everything with this format, and compare" (2026-03-31 team-desiree). Naming supplies the data that keeps the Creative Tree usable.

Iteration discipline#

Winners get iterated immediately and repeatedly: new hooks on proven bodies ("test multiple hooks and keep the same body", 2026-07-23 team PM), shorter cuts, POV reframes, localized versions. Volume follows signal, not the other way round. The working benchmark he endorses: roughly one live ad per $10k of monthly spend, weighted toward hit rate over raw output: "you can make 100 ads, but if none of them work, you made 100 ads for no fucking reason" (2026-07-03 client-luke).

The AI line#

AI amplifies craft; it does not replace it. The rule as given to the video team: if a concept requires AI to exist and would look AI-made, skip it; if AI fills gaps invisibly (B-roll, asset variants, expansions) or is deliberately, obviously AI (animation, characters), it earns its place. "We're at that inflection point... for it to qualify, it's got to not look like AI" (2026-08-10 team-creative-workflow). Real founder footage beats generated content: "a founder video will fucking smash any AI bullshit you can make" (2026-03-31 peer-manny). Market evidence he cites for the volume thesis: IM8 grew to roughly $100M revenue in under a year largely on high-volume, low-polish AI creative with a premium site behind it (2026-06-29 team-creative-review). Internal figure, not for publication.

The defense: ads don't build the brand#

When brand owners push back on unpolished ads (a recurring VIBAe founder objection: "don't you think they make our brand look cheap?"), the standing answer: "the brand isn't built on the ads. It's everything that happens after they click... the ad is just to get people's attention" (2026-07-01 client-vibae-monthly). Supporting points: buyers convert, wholesale partners report ad-driven foot traffic, and polish caps scale ("we wouldn't be able to spend 600,000 euros a month on Meta and still get the same performance"). The concession that keeps trust: tighten quality gates so genuinely off-brand work never ships, without slowing the flow.

Related: working method, receipts standard, content strategy, spoken voice profile.